With development of a photoelectric gathering technology and an increasingly growing requirement for a high-definition digital video, a video data volume is growing, finite heterogeneous transmission bandwidth and diversified video applications continuously impose a higher requirement for video encoding efficiency, and work of formulating the High Efficiency Video Coding (HEVC) standard starts as needed.
A basic principle of video encoding and compression is to eliminate redundancy to an utmost extent using a correlation between a space domain, a time domain, and a code word. Currently, in a popular practice, the video encoding and compression are implemented by performing steps such as prediction (including intra-frame prediction and inter-frame prediction), conversion, quantization, and entropy encoding using a block-based hybrid video encoding framework. In an intra-frame prediction technology, redundant information of a current image block is eliminated using space pixel information of the current image block, to obtain a residual. In an inter-frame prediction technology, redundant information of a current image block is eliminated using pixel information of a coded or decoded image adjacent to a current image, to obtain a residual. This encoding framework demonstrates strong vitality. This block-based hybrid video encoding framework continues to be used in HEVC as well.
In 3D video encoding and decoding, to reflect depth information of an object, encoding and decoding need to be performed on a depth map. Data of a depth map has some signal characteristics different from data of a texture map. The most prominent signal characteristic is that there are many smooth areas with similar depth values, and in many cases, a same depth value may be used for sampling points in a smooth area. Therefore, an encoding and decoding algorithm based on a single depth intra-frame mode (SDM) is introduced for an image block of a depth map. A core of the SDM is to simplify reconstruction of a current image block, and one depth value may be used to represent a depth value of an entire smooth area of the current image block.
In a traditional SDM-based encoding and decoding algorithm, at least five adjacent prediction sampling points around a current image block need to be detected according to a specified detection sequence, to obtain a corresponding depth value and write the corresponding depth value into a sample candidate set, that is, in a traditional mechanism, the sample candidate set is constructed according to detection results of the at least five adjacent prediction sampling points. Because the at least five adjacent prediction sampling points need to be referred to when the sample candidate set is constructed, encoding and decoding complexity of the traditional mechanism may be relatively high, and relatively large storage space is needed.